I had a first discussion on irc about BeOS. Please read it, it has some interesting ideas about development directions :) naish = Michael Wulff Nielsen marcus_o = Marcus Overhagen Time is german local time (sorry) [19:14] <@naish> Hi [19:14] <marcus_o2> hello [19:14] <@naish> Welcome back [19:14] <marcus_o2> to bads there are only two oh us here now [19:14] <marcus_o2> Are you from the Netherlands? [19:15] <@naish> No I am from Denmark, what makes you ask? [19:16] <marcus_o2> Because of the time, I'm from Germany [19:16] <marcus_o2> And in USa now it's what time? [19:16] <marcus_o2> Evereybody still at work I guess [19:16] <@naish> Ohh, I see. Well later on more people will show up. USA is 8 hours behind us [19:17] <marcus_o2> I'm very disappointed that Be did not release their CAPI and AVM B1 driver as sample code about a year ago [19:18] <marcus_o2> With sound drivers it did work, there are many sound card drivers at BeBits [19:18] <@naish> Well, I am disappointed too, but it is in the incredible stupidity of Be. ISDN drivers are just part of the problem. [19:19] <marcus_o2> But as there was no interface, and since BONE was promised to appear very soon, nobody did do anything about it [19:19] <marcus_o2> I think the problem is that most useres in europe do use internal ISDN cards, and don't want to by a new one [19:19] <@naish> As they mostly do. Well I have started hacking on a proof-of-concept app_server which can run on top of a linux kernel using the framebuffer. [19:20] <marcus_o2> well, will it be possible to run BeOS apps without recompiling? [19:21] <marcus_o2> I think you know I don't like the approach using linux, but we need to keep the users [19:21] <@naish> Nope, I don't care about being backwards compatible. I want a new kernel and a new beos. That is my only major problem with the openbeos team, but I have decided to go with whatever decision the team makes. [19:22] <marcus_o2> I also do not like beeing backwards compatible, but I fear that we will loose allmost any current Be user if we do as you want to do [19:23] <@naish> Perhaps, but a good kernel with a lot of drivers will allow us to attract new users. Linux users will come flocking in the thousands when we provide with an OT/OD alternative to KDE [19:23] <marcus_o2> psosible [19:23] <marcus_o2> possible [19:24] <@naish> But I admit that replacing parts one at the time is possible and somewhat attractive. But reverse engineering every systemcall in the be kernel and reimplementing libbe.so is an incredibly big task. With linux we get a kernel and a glibc.so right away [19:25] <marcus_o2> Yes, true. But on the other hand, with Linux we get Linux, if you understand what I want to say [19:25] <@naish> Not really, you may have to clarify that.. sorry [19:26] <marcus_o2> If you use the linux kernel, you loose most of the modularity the BeOS kernel did rpovide [19:27] <@naish> May I ask how much you know about the linux kernel? No offence intended [19:27] <marcus_o2> you have then another linux kernel, and are about to create something like a new KDE [19:28] <marcus_o2> I don't know much, but I once installed it, and I donwloaded all the recent sourcec to get familiear with the concept [19:28] <@naish> There would be a fundemental difference between KDE and Linux/BeOS. No X-server, no bloatware windowmanager, no illogical UI... The linux kernel is pretty modular and very stable/fast. [19:29] <marcus_o2> I also had a look at the NewOS kernel, written by the former Be engineer Travis Geiselbrecht, it's nice, but very unevolved [19:30] <marcus_o2> AtheOS is nearly the same, except that the guy did a easy implementation off an app_server clone [19:30] <@naish> My point ecxatly, the linux kernel is very evolved. Provides full posix compatibility, has support for a dozen filesystems, has USB/SCSI. And is being actively worked on. [19:31] <marcus_o2> I don't know if we better use Linux, seriously, I don't expect this to help us creating a BeOS. It will remain a Linux :) [19:31] <@naish> AtheOS is completely unsuitable because of the author kurt's attitude towards beos. Don't get me wrong, he is a great guy, and the best programmer I have ever hacked code with. [19:31] <marcus_o2> I also don't like the attitude he shows on his homepage [19:32] <marcus_o2> He has clearly stolen the BeOS idears & concepts, and denying this is not making him look very good [19:32] <@naish> I think using a linux kernel would make a new hybrid, something more evolved than beos and much more usable than linux [19:33] <marcus_o2> I agree that writing a complete new kernel will take probably much too long [19:34] <@naish> Time we don't have. Hacking a new appserver is possible and a new storage kit. Then we could compile opentracker. Upgrade to the newest stable gcc and go from there. [19:34] <marcus_o2> We will also need the complete Media Kit [19:34] <marcus_o2> then, i fear [19:35] <@naish> Of course, but it could come second [19:35] <@naish> What do you fear? [19:37] <marcus_o2> I think replacing some of the XXX_server of BeOS and at the same time writing / porting a kernel that can load all existing BeOS drivers and stays binary compatible with BeOS is the best way to start. We may break compatiblility later, but if we break today, we are lost, I fear. [19:38] <marcus_o2> I think we will need current users & developers & software, else we and up as just another open source OS project [19:38] <@naish> why? most people could stay with r5 while we mature the new beos [19:39] <marcus_o2> This is what I see as the problem. The won't stay with an no longer developed BeOS, since they already waited too long [19:39] <marcus_o2> And they will se a replacement OS, on which they can't run there programs, and which is not possible of reading there files on BFS partition [19:40] <@naish> You might be right, but I fear that it would be almost impossible to replace the Be kernel by reverse engineering. [19:40] <marcus_o2> not as the way to continue [19:40] <marcus_o2> I don't want to do much rewerse engineering [19:40] <marcus_o2> We need to know which functions for example libnet.so provides [19:40] <marcus_o2> then we create an replacement [19:41] <marcus_o2> most driver interfaces are documented in form of sample code [19:41] <marcus_o2> so you should be able to do it [19:41] <marcus_o2> I don't want to revere engineer the kernel [19:42] <@naish> perhaps. but the new BeOs would of course have a BFS driver, that would be a requirement [19:42] <marcus_o2> ok, but I don't think this will be enough [19:42] <@naish> And you would do that with libbe.so, try checking out the amount of functions defined in libbe.so [19:43] <marcus_o2> yes, in the end phase, we would have to replace this one, too [19:43] <marcus_o2> how many functions does it export? a few hundret? should not be that much of a problem [19:43] <@naish> I don't know if that is the most insane or the bravest thing I have ever heard. [19:44] <@naish> Probably around 2 -> 3 thousand if my guess is correct. [19:44] <marcus_o2> I'm not sure, too. It really depends on how many people we get [19:44] <marcus_o2> I don't think it are more than thousend. Will be much less [19:45] <marcus_o2> isn't libbe.so available as source code, since they needed to provide it as it *IS* the glibc? [19:45] <@naish> Unfortunately I don't have be partition right now, since my new AMD won't boot BeOS (another: We need a new kernel fast) [19:46] <@naish> No we don't have libbe.so, there is no need to release the source code. [19:47] <marcus_o2> Ok, I'm wrong [19:47] <marcus_o2> libroot.so system/lib Standard C library, POSIX-style system calls, basic C++ runtime support. This library is required for almost anything except the simplest of shared libraries and add-ons. [19:47] <marcus_o2> libbe.so system/lib Most of the BeOS C++ Kits. The Interface Kit and the Application Kit are the two biggies in this shared library. Required for anything that wants to present a user interface on BeOS. [19:48] <marcus_o2> You are correct. May be a few thound of functions [19:48] <@naish> An enourmous task at best, impossible at worst. [19:48] <marcus_o2> BTW, this is from http://www.jump.net/~hplus/dr9.html [19:49] <marcus_o2> It's possible to do [19:50] <marcus_o2> The old homepage of John Wtte: http://www.zilker.net/~hplus/index2.html [19:51] <marcus_o2> From the time when he was still a Be Employe [19:51] <@naish> Sure, but will we want to do it, perhaps we should take Be as an example and write a new os. [19:52] <marcus_o2> This has been too many times. Do you know for example, moment, i grabbing for the link [19:53] <marcus_o2> This has been DONE too many times. Do you know for example http://www.skyos.org/index2.html ? [19:55] <marcus_o2> It's always the same problem, no user base. Creating a new open Source OS incomplatible with BeOS will be a huge problem. You won't get the users over onto your os. On the other hand, rewriting a compatible BeOS may too much word and not possbile at all, I know this [19:55] <marcus_o2> may be too much WORK! I wanted to say :) [19:57] <@naish> okay, I think my decision is the right one. Whatever the openbeos team decides I will back them [19:58] <marcus_o2> Ok, I think my decision is the right on, too :-) Perhaps we should post this discussion on the mailinglist might be interesting for others, do you agree [19:59] <@naish> Yes, it would, will you do it? [19:59] <marcus_o2> Ok, I will do it [19:59] <@naish> Thanx. So where do we start? [20:01] <marcus_o2> Don't know, too bad. Both ways to start this project may most probably not be the right ones :) [20:02] <marcus_o2> But please have a look at the Palm, Inc stock: http://finance.yahoo.com/q?s=palm&d=c&t=1y> [20:02] <@naish> Well, lets think? What is the smallest corner of beos that would need reimplementing [20:02] <@naish> OUCH [20:02] <marcus_o2> Looks very similar to Be Inc. [20:03] <marcus_o2> Networking? Sound? Video? OpenGl? Booting on P4 and AMD? [20:03] <@naish> INcredibly. But the sell a product and they have cash [20:05] <@naish> Sorry if I am a bit distracted, but I am reconfiguring my home network to accomodate a BeOS laptop to develop on [20:05] <marcus_o2> Be Inc also was selling a product. But it didn't have enogh cash. But you don't know which Problems Palm will face in a year. I think it's the correct decision to do something without Palm, although I would ike them to open source BeOS, o [20:06] <marcus_o2> or to put out BONE & OpenGL to enjoy the user base [20:06] <@naish> med too, but lets just do it. there isn't any companies left to sue us anyway [20:07] <marcus_o2> I agree, legal problems will most likely not hapen, as nobody is left who interested in suing us [20:07] <@naish> We should host all files in russia with backups all over the world ;) [20:08] <marcus_o2> We should simply stay away from DeCSS to be save :) [20:09] <@naish> Probably [20:10] <marcus_o2> I think there was a report on benews some time ago that newer P4 processors do work with BeOS, and tht only the old ones failed. So what's happending with your AMD? I assumed ever one would work [20:10] <@naish> It won't boot, freezes in the boot screen [20:10] <marcus_o2> did you enable console debug output? [20:11] <@naish> No, now I use my thinkpad a20m it never fails [20:12] <marcus_o2> Ok, fine :) [20:13] <marcus_o2> Did you joint the BlueOS team? [20:13] <@naish> I have to change cables. need to get the laptop online. And no I will stay away from blueos regards Marcus